[riot-users] possible hwtimer bug

Marco Ziegert ziegert at inf.fu-berlin.de
Sun Apr 14 13:02:07 CEST 2013


Hello Aleksandr,

as the name says hwtimer is hardware dependent - can you tell us which MCU you are using? 

If you are programming for the STM Cortex-M3/M4 - only 16-Bit Timer are implemented for hwtimer. 32-Bit Timer are very rare and may be used for more important tasks than waiting for longer periods. E.g. one 32-Bit Timer is temporarily used for hwtimer_now(). The longest period you can wait with a 16-Bit Timer are 54-56 seconds as far as I remember the implementation. For longer periods use the RTC alerts.

Best regards,
Marco Ziegert

Am 14.04.2013 um 12:39 schrieb Aleksandr Mikoff <sir.enmity at gmail.com>:

> Dear developers and users,
> Today i discovered a very strange behaviour of scheduler. When i perform hwtimer_wait(x) function, where x - time to wait, if x < critical_value then the function is finished correctly, but if x>critical_value, then the function never wakes up. The critical_value varies and depends on time of excectuion of sched_set_status function.
> I have the following output from scheduler when the  function doesn`t wakes up:
> thread_wakeup: Trying to wakeup PID 1..
> status = 4875(or = 64)
> thread_wakeup: Thread is not sleeping! 
> I tried to catch the function, which change the pid_status value, but nothing came out.
> Let me know, if you have any ideas, thank you.
> 
> -- 
> Best regards,
> Aleksandr Mikov
> _______________________________________________
> users mailing list
> users at riot-os.org
> http://lists.riot-os.org/mailman/listinfo/users


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.riot-os.org/pipermail/users/attachments/20130414/788720cc/attachment.html>


More information about the users mailing list